Identification of Differential Compositions of Aqueous Extracts of Cinnamomi Ramulus and Cinnamomi Cortex

Cinnamomi ramulus (CR) and Cinnamomi cortex (CC), both sourced from Cinnamomum cassia Presl, are commonly used Chinese medicines in the Chinese Pharmacopeia. However, while CR functions to dissipate cold and to resolve external problems of the body, CC functions to warm the internal organs. To clarify the material basis of these different functions and clinical effects, a simple and reliable UPLC-Orbitrap-Exploris-120-MS/MS method combined with multivariate statistical analyses was established in this study with the aim of exploring the difference in chemical compositions of aqueous extracts of CR and CC. As the results indicated, a total of 58 compounds was identified, including nine flavonoids, 23 phenylpropanoids and phenolic acids, two coumarins, four lignans, four terpenoids, 11 organic acids and five other components. Of these compounds, 26 significant differential compounds were identified statistically including six unique components in CR and four unique components in CC. Additionally, a robust HPLC method combined with hierarchical clustering analysis (HCA) was developed to simultaneously determine the concentrations and differentiating capacities of five major active ingredients in CR and CC: coumarin, cinnamyl alcohol, cinnamic acid, 2-methoxycinnamic acid and cinnamaldehyde. The HCA results showed that these five components could be used as markers for successfully distinguishing CR and CC. Finally, molecular docking analyses were conducted to obtain the affinities between each of the abovementioned 26 differential components, focusing on targets involved in diabetes peripheral neuropathy (DPN). The results indicated that the special and high-concentration components in CR showed high docking scores of affinities with targets such as HbA1c and proteins in the AMPK–PGC1–SIRT3 signaling pathway, suggesting that CR has greater potential than CC for treating DPN.


Introduction
Cinnamomi ramulus (CR, Guizhi in Chinese) and Cinnamomi cortex (CC, Rougui in Chinese), are, respectively, the dried twigs and bark of Cinnamomum cassia Presl. They are documented in the Chinese Pharmacopoeia, which has a long history and comprises all well-known Chinese medicines with a high clinical value [1]. CR functions to dissipate wind-cold and resolve external problems of the body; it traverses the arms and warms the meridians. It is known as the first medicine of classic prescription and is widely used in the Treatise on Cold Pathogenic and Miscellaneous Diseases [2]. In contrast, CC functions to warm the internal organs, nourish and warm kidney-yang, dispel cold, and relieve pain. Different pharmacological effects result from variant chemical ingredients. It can be vaporizer temperature was 350 °C and the positive ion spray voltage was 3.5 kV. The above data are shown in Table S1.

Identification of the Constituents of CR and CC
Qualitative analysis of the chemical constituents of the aqueous extracts of CR and CC was performed using UPLC-Oribtrap-Exploris-120-MS/MS, in both in the positive and negative modes. In the qualitative analysis, in order to evaluate the stability of the instrument, mixed standards (QC) were repeated for 8 times. The RSDS of the retention times and intensities in QC samples were all less than 5%. Typical total ion chromatograms for CR and CC, both in negative ion mode are shown in Figure S1. A comparison of retention times, accurate mass, fragmentation patterns, and a comparison with an online database, as well as referencing to the related literature to validate the data, preliminarily identified 58 compounds (Table 1, Figures S2 and S3); these included nine flavonoids, 23 phenylpropanoids and phenolic acids, two coumarins, four lignans, four terpenoids, 11 organic acids and five other components. Among these compounds, eight compounds were identified by comparing their retention times, characteristic molecular ions, and fragment ions to those of the standards ( Figure S4). Significantly, among these compounds, 26 significant differential compounds were identified statistically: the results showed that the peak areas of flavonoids and flavonoid glycosides, and phenylpropanoids except cinnamaldehyde and cinnamyl alcohol in CR were higher than in CC, and the peak areas of terpenoids and organic acids in CC were higher than in CR.

Identification of Phenylpropanoids and Phenolic Acids
Phenylpropanoids and phenolic acids are major bioactive constituents in CR and CC [22]. Trans-cinnamic acid, trans-cinnamaldehyde, cinnamyl alcohol and 2-methoxycinnamic acid were shown to be the main phenylpropanoids. In the literature, the diagnostic ions at m/z 163/165, 151/153 and 135/137 105/107 were indicators of phenylpropanoids and phenolic acids [23]. However, in addition to the above diagnostic ions reported in the literature and standards, our statistical analysis found that fragment ions at m/z 105/107 and 123/125 can also be diagnostic ions [23].
With the transfer of electrons, the carbonyl group was also prone to cleavage and the loss of CO and CH2 to form fragment ion peaks [24,25]. As a result, when compared to the diagnostic ions described above, the mass spectra of the standards, fragmentation information in the database and previously reported fragmentation spectra, a total of 23 phenylpropanoids and phenolic acids were identified in CR and CC.  (Table 1), which were in general agreement with the literature and data available for cinnamic acid; accordingly, the compound was presumed to be trans-cinnamic acid [26]. The MS 2 spectrum of compound 45 is shown in Figure S5 and the possible cleavage process of the positive ions is shown in Figure 1.    (Table 1), which was consistent with the cleavage fragment of trans-cinnamaldehyde in the literature and database: the compound was, therefore, presumed to be trans-cinnamaldehyde [57]. The MS 2 spectrum of compound 48 is shown in Figure S6, and the possible cleavage process of the positive ions was shown in Figure 2.  (Table 1), which was consistent with the cleavage fragment of trans-cinnamaldehyde in the literature and database: the compound was, therefore, presumed to be trans-cinnamaldehyde [57]. The MS 2 spectrum of compound 48 is shown in Figure S6, and the possible cleavage process of the positive ions was shown in Figure 2.

Identification of Flavonoids
The RDA cleavage of the C-ring was the main cleavage of flavonoids, resulting in fragment ions and the loss of a series of neutral small molecules such as H2O, CO2, and CO by energy collisions under certain mass spectrometric conditions [25,58]. Flavonoid oxyglycosides generally lost sugar first, and then cleaved according to the cracking law of flavonoid skeleton structure [26,59]. Combined with the mass fragments information in the literature and standards, the ionic fragments at m/z 291/289, 303/301, 287/285 were selected, respectively, as the diagnostic ion for epicatechin-type, quercetin-type, and kaempferol-type [23]. The cracking of flavonoids and dihydroflavones was very similar, for example, epicatechin (11) and quercetin (32) demonstrated the cracking laws of epicatechin-type and quercetin-type, as shown in Figures 3 and 4, the MS 2 spectrum of epicatechin (11) is shown in Figure S7.

Identification of Flavonoids
The RDA cleavage of the C-ring was the main cleavage of flavonoids, resulting in fragment ions and the loss of a series of neutral small molecules such as H 2 O, CO 2 , and CO by energy collisions under certain mass spectrometric conditions [25,58]. Flavonoid oxyglycosides generally lost sugar first, and then cleaved according to the cracking law of flavonoid skeleton structure [26,59]. Combined with the mass fragments information in the literature and standards, the ionic fragments at m/z 291/289, 303/301, 287/285 were selected, respectively, as the diagnostic ion for epicatechin-type, quercetin-type, and kaempferol-type [23]. The cracking of flavonoids and dihydroflavones was very similar, for example, epicatechin (11) and quercetin (32) demonstrated the cracking laws of epicatechintype and quercetin-type, as shown in Figures 3 and 4, the MS 2 spectrum of epicatechin (11) is shown in Figure S7.  (Table 1), which was consistent with the cleavage fragment of trans-cinnamaldehyde in the literature and database: the compound was, therefore, presumed to be trans-cinnamaldehyde [57]. The MS 2 spectrum of compound 48 is shown in Figure S6, and the possible cleavage process of the positive ions was shown in Figure 2.

Identification of Flavonoids
The RDA cleavage of the C-ring was the main cleavage of flavonoids, resulting in fragment ions and the loss of a series of neutral small molecules such as H2O, CO2, and CO by energy collisions under certain mass spectrometric conditions [25,58]. Flavonoid oxyglycosides generally lost sugar first, and then cleaved according to the cracking law of flavonoid skeleton structure [26,59]. Combined with the mass fragments information in the literature and standards, the ionic fragments at m/z 291/289, 303/301, 287/285 were selected, respectively, as the diagnostic ion for epicatechin-type, quercetin-type, and kaempferol-type [23]. The cracking of flavonoids and dihydroflavones was very similar, for example, epicatechin (11) and quercetin (32) demonstrated the cracking laws of epicatechin-type and quercetin-type, as shown in Figures 3 and 4, the MS 2 spectrum of epicatechin (11) is shown in Figure S7.    (Table 1, Figure 3), This was consistent with the cleavage fragment of epicatechin in the literature and standards; accordingly, the compound was presumed to be epicatechin [59].  Figure 4). This was consistent with the cleavage fragment of quercetin in the literature [59] and standards, so the compound was presumed to be quercetin.

Identification of Coumarins
The main cleavage of coumarins included the losses of OH, CH3 and CO2 from characteristic substituents and the continuous neutral loss of CO from the pyran ring [58,60]. The fragmentation rule for the coumarin glycosides was similar to that for coumarins, except for the initial losses of sugar moieties.
As  (Table  1). This was consistent with the cleavage fragment of coumarin in the literature [61,62] and standard, so 31 was presumed to be coumarin. The MS 2 spectrum of compound 31 was shown in Figure S8, and the possible cleavage process of the positive ions is shown in Figure 5.    Figure 3), This was consistent with the cleavage fragment of epicatechin in the literature and standards; accordingly, the compound was presumed to be epicatechin [59].
Compound 32 had the molecular formula C 15 Figure 4). This was consistent with the cleavage fragment of quercetin in the literature [59] and standards, so the compound was presumed to be quercetin.

Identification of Coumarins
The main cleavage of coumarins included the losses of OH, CH 3 and CO 2 from characteristic substituents and the continuous neutral loss of CO from the pyran ring [58,60]. The fragmentation rule for the coumarin glycosides was similar to that for coumarins, except for the initial losses of sugar moieties.
As  (Table 1). This was consistent with the cleavage fragment of coumarin in the literature [61,62] and standard, so 31 was presumed to be coumarin. The MS 2 spectrum of compound 31 was shown in Figure S8, and the possible cleavage process of the positive ions is shown in Figure 5.  Figure 4). Thi was consistent with the cleavage fragment of quercetin in the literature [59] and standards so the compound was presumed to be quercetin.

Identification of Coumarins
The main cleavage of coumarins included the losses of OH, CH3 and CO2 from char acteristic substituents and the continuous neutral loss of CO from the pyran ring [58,60] The fragmentation rule for the coumarin glycosides was similar to that for coumarins except for the initial losses of sugar moieties.
As  (Table  1). This was consistent with the cleavage fragment of coumarin in the literature [61,62 and standard, so 31 was presumed to be coumarin. The MS 2 spectrum of compound 31 was shown in Figure S8, and the possible cleavage process of the positive ions is shown in Figure 5.  Identification information of other compounds were detailed in Sections S1-S7 of Supplementary Materials.

Statistical Analysis
The MS data from different sources of CR and CC were imported into Compound Discoverer 3.2 qualitative analysis software for normalization and export, with a total of 1460 positive and negative ion fragments. The processed data were imported into SICMA 14.1 software for statistical analysis, resulting in the designation of 37 discrepant compounds. Analysis of variance was performed on the discrepant compounds using GraphPad Prism 7.0 software, resulting in 26 statistically significant (p < 0.05) discrepant compounds.

Principal Component Analysis (PCA)
PCA analysis can clearly explain the correlation between a large number of variables and a small number of principal components with less data loss; this can effectively reduce the dimensionality of the multidimensional raw data. The results showed that the CR and CC samples could be clearly separated and classified into one category, indicating that there were differences in the chemical compositions of CR and CC. When statistical analysis was used, the QC samples should be included to evaluate the instrument stability. The results of the PCA analysis are shown in Figure S9.

Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA)
The difference in the chemical compositions of CR and CC was further studied using supervised OPLS-DA; the score diagram was shown in Figure S10. The results indicated that CR and CC can be clearly separated, which was consistent with the PCA results. The model fitting was also validated by setting the number of tests to 200. The R 2 and Q 2 points on the left of the data were lower than the original values on the right, and the regression line at the Q 2 intersected the vertical axis below the original point with a negative nodal increment. The results of the model validation were R 2 Y = 0. 999 and Q 2 = 0.952 (R 2 in the validation results represented the matching of the model. The closer R 2 to 1, the more complete the model was in describing the data; the closer Q 2 to 1, the higher the predictive ability of the model) ( Figure S11). Model evaluation results indicated that the model had robustness and no over-fitting. The commonly used variable importance for the project (VIP) was used to estimate for quality difference markers between groups. Variables with a VIP > 1 was generally considered to be meaningful to the model, the greater the VIP value was, the greater the contribution of the variable. Therefore, a VIP > 1 was selected to analyze the overall variability of CR and CC to find differential markers ( Figure  S12). A total of 627 characteristic peaks was determined for the differential components (VIP > 1) that contributed significantly to the classification of CR and CC (Table S2). A total of 37 differential compounds with VIP > 1 was identified based on the analysis of compound retention times, accurate relative molecular masses, and cleavage information. Information on compounds with VIP > 1 were shown in Table S2. The peak areas of these 37 compounds were subjected to t-tests using GraphPad Prism7.0 software, resulting in 26 chemical components with VIP > 1 and p < 0. 05, including compounds 1-3, 5, 8, 16,  17, 19, 24, 26, 27, 31-36, 38, 40-42, 45, 46, 48, 53 and 58, These comprised two organic acids, six phenolic acids, seven phenylpropanoids, eight flavonoids, two terpenoids and one coumarin. According to the peak areas and secondary fragment information, of the 26 significantly different compounds, 6 (19, 32, 33, 38, 41 and 53) were specific to CR, and 3 (8, 26, and 36) were specific to CC.

Semi-Quantitative Analysis of CR and CC
A semi-quantitative analysis was further carried out to compare the intensity trends of CR and CC by calculating the relative peak areas of 26 compounds in 16 samples. The boxlots could visually observe the content changes between chemical compositions. The peak areas data for the 26 compounds in CR and CC were exported, and the boxplots produced by GraphPad Prism 7.0 software were further used to compare the relative content of the compounds in CR and CC, as shown in Figure 6. box-lots could visually observe the content changes between chemical compositions. The peak areas data for the 26 compounds in CR and CC were exported, and the boxplots produced by GraphPad Prism 7.0 software were further used to compare the relative content of the compounds in CR and CC, as shown in Figure 6. Accordingly, the peak areas of flavonoids and flavonoid glycosides, and phenylpropanoids except trans-cinnamaldehyde in CR were higher than in CC, and the peak areas of terpenoids and organic acids in CC were higher than in CR.

Method Validation
The relative standard deviations (RSDs) for the precision, stability, and repeatability investigations into of coumarin, cinnamyl alcohol, cinnamic acid, dimethoxy cinnamic acid and cinnamaldehyde were all < 5% ( Table 2), indicating that the method had good precision, repeatability, and recovery. Additionally, a recovery range of 97−101% (RSD < 4%) indicated its high recovery and reliability (Table S3).  Accordingly, the peak areas of flavonoids and flavonoid glycosides, and phenylpropanoids except trans-cinnamaldehyde in CR were higher than in CC, and the peak areas of terpenoids and organic acids in CC were higher than in CR.

Method Validation
The relative standard deviations (RSDs) for the precision, stability, and repeatability investigations into of coumarin, cinnamyl alcohol, cinnamic acid, dimethoxy cinnamic acid and cinnamaldehyde were all <5% ( Table 2), indicating that the method had good precision, repeatability, and recovery. Additionally, a recovery range of 97−101% (RSD < 4%) indicated its high recovery and reliability (Table S3).

Quantitative Determination of the Major Constituents in CR and CC Using HPLC
The established HPLC analysis method was subsequently used to determine the representative components in eight batches of CR and CC products. In the case of quantitative analysis, all the samples (eight batches of CR and CC) were extracted three times and analyzed by HPLC. The RSD value of the concentrations of these five standards were less than 5%. The HPLC chromatograms were shown in the Figure S13. The contents of the five compounds are summarized in Table 3, based on their respective calibration curves. The results showed that there were significant differences in the composition content of CR and CC. Among them, the content of trans-cinnamaldehyde in CC was about twice that in CR, and the content of trans-cinnamic acid in CC was similar to that in CR; the contents of coumarin, cinnamyl alcohol and 2-methoxycinnamic acid in CC were significantly higher than those in CC. We, therefore, believe that these five components play a key role in the different efficacies of CR and CC.

Cluster Analysis
The original data of our compound content was imported into Lianchuan Biotechnology's advanced heat map statistics software, and the original data was normalized by Z-score to obtain the result ( Figure S14). As our statistics result indicated, CR and CC were each clustered into one group. Five compounds were screened out and could be used as chemical markers for distinguishing CR and CC; Among the five, two markers including trans-cinnamaldehyde and cinnamyl alcohol (VIP value is greater than the other three compounds) made greater contributions to sample grouping than the other three ones.

Molecular Docking
The different screened constituents were docked to PGC1α (PDB ID: 1XB7), SIRT3 (PDB ID: 4BN4) and AMPK (PDB ID: 4CFF). CDOCKER_INTERACTION_ENERGY ≤ −5.0 kJ/mol was used to produce a better binding ability for the compounds and proteins. The results were shown in Table S4 and Figure 7.
The results showed that compounds 40, 34, 17, 26 and 27 had the highest binding energy with Glycosylated Hemoglobin, Type A1C (HbA1c); compounds 1, 32, 16 and 19 had the highest binding energy with peroxisome proliferator-activated receptor-gamma coactivator-1alpha (PGC1α); compounds 24, 2, 45, 42 and 3 had the highest binding energy with silent information regulator protein 3 (SIRT3); and compounds 40, 33, 34, 27 and 32 had the highest binding energy with AMP-activated protein kinase (AMPK). Among these compounds, only 2, 3, 17, 26 had a higher content in CC. These results demonstrated that the special and high-concentration components in CR showed high docking scores of affinities with targets such as HbA1c and the proteins in the AMPK-PGC1-SIRT3 signaling pathway, suggesting the greater potentials for CR in treating DPN than for CC. The compounds with high protein binding energy were phenylpropanoid (1, 16, 42, 45) and flavonoids (19, 26,  27, 40).
The results showed that compounds 40, 34, 17, 26 and 27 had the highest binding energy with Glycosylated Hemoglobin, Type A1C (HbA1c); compounds 1, 32, 16 and 19 had the highest binding energy with peroxisome proliferator-activated receptor-gamma coactivator-1alpha (PGC1α); compounds 24, 2, 45, 42 and 3 had the highest binding energy with silent information regulator protein 3 (SIRT3); and compounds 40, 33, 34, 27 and 32 had the highest binding energy with AMP-activated protein kinase (AMPK). Among these compounds, only 2, 3, 17, 26 had a higher content in CC. These results demonstrated that the special and high-concentration components in CR showed high docking scores of affinities with targets such as HbA1c and the proteins in the AMPK-PGC1-SIRT3 signaling pathway, suggesting the greater potentials for CR in treating DPN than for CC. The compounds with high protein binding energy were phenylpropanoid (1, 16, 42, 45) and flavonoids (19, 26, 27, 40).

Preparation of Sample Solutions
Each CR and CC sample was weighed (1 g) accurately and extracted under reflux with 50 mL ultrapure water for 2 h. They were left to reach room temperature; water was then added to compensate for the weight loss of the extraction solution. A volume of 1 mL of the extraction solution was diluted with methanol to 2 mL; this was filtered through a 0.22 µm nylon filter membrane and centrifuged at 12,000 rpm for 15 min for UPLC-MS analysis. A volume of 1 mL of the extraction solution was directly filtered and centrifuged for HPLC quantitative analysis. Mixed standard solutions (trans-cinnamaldehyde, trans-cinnamic acid, epicatechin, 2-methoxycinnamic acid and quercitrin) were used as QC samples.

Preparation of Reference Solutions
An appropriate amount of the reference materials of coumarin, cinnamyl alcohol, cinnamic acid, 2-methoxycinnamic acid and cinnamaldehyde were measured and weighed precisely; methanol was added to dissolve them, and 1.0331, 1.8667, 0.6667, 1.1333 and 3.4 mg/L, respectively, were prepared for standby.

UPLC-MS Method for Qualitative Analysis
MS data was acquired in fast chromatography MS 2 mode, the mass spectrometer parameters were set as follows: The ESI was used in negative ion mode (ESI−) and in positive ion mode (ESI+). The following settings were used: the spray voltage was 2.5 kV(−) and 3.5 kV(+); the UHPLC-MS/MS mode was applied with an Orbitrap resolution of 120,000 for full-MS and 15,000 for dd-MS 2 ; the isolation window (m/z) was 2; the RF Lens% was 70; the sheath gas pressure was 45 Arb; the auxilliary gas pressure was 15 Arb; the sweep gas pressure was 0 Arb; the capillary temperature was 320 • C; vaporizer temperature was 350 • C; the scanning range was m/z 80~800; the stepped normalized collision energies (NCE) were 15, 30 and 45 eV.

HPLC Method for Quantitative Analysis
The major active components including coumarin, cinnamyl alcohol, trans-cinnamic acid, 2-methoxycinnamic acid and trans-cinnamaldehyde were used for quantitative analysis.
To evaluate linearity, a series of standard solutions with appropriate concentrations were obtained by diluting standard compounds with methanol. The calibration curves were drawn with the quality of the reference substance as the abscissa (X) and the peak area as the ordinate (Y), they showed a good linear relationship (r ≥ 0.999) within the test ranges. The same sample solutions of CR and CC were continuously injected to verify the precision of the instrument. The same sample solutions were injected, separately, 0, 2, 4, 8, 12, and 24 h separately to check the stability of the test solution. Six sample solutions were prepared independently to verify the repeatability of the method. The accuracy of the method was evaluated by a recovery test. Recovery (%) = {[Found − (Original sample + Add)]/Add} * 100.

LC-MS Data Processing and Statistics
The UPLC-Orbitrap-Exploris-120-MS data were acquired by the Trancefinder 5.1 software; the UHPLC-MS/MS mode was applied with an Orbitrap resolution of 120,000 for full-MS and 15,000 for dd-MS 2 . The date was then analyzed using the Xcalibur3.0 software. Each raw data processed using Compound Discoverer 3.2 qualitative analysis software following a specific workflow ( Figure S15). The screening steps of the target compound were: All ions presenting a signal over 5 times the background noise and a peak intensity over 10 5 were taken into account to create the extracted ion chromatogram (EIC). MS and MS 2 spectra were then used to identify ions by searching the mzCloud, mzValut, Chemspider and mass Lists databases. Then select the matching compounds with a tolerance of 5 ppm and more than 60% of the database matching degree as our preliminary identified compound. Finally, the compounds were reconfirmed by combined with the parent ion and the MS2 fragment ions extracted from their original data with those in the relevant literature. Additionally, the compounds were all rechecked by searching literature materials to exclude the non-natural products. The peak area of fragment ions was statistically analyzed to determine the statistically significant difference components, and then the marker compounds were clustered for the purpose of distinguishing CR and CC.
To evaluate relationships on the basis of similarities or differences between groups of multivariate data, multivariate analyses (PCA and OPLS-DA) were performed using SICMA 14.1. PCA results were displayed in the form of score plots. OPLS-DA was conducted using class information as the Y-variable; the results were shown in the form of score plots. The contribution of variables to the analysis was explained using variable importance in projection (VIP) scores. VIP scores are a weighted sum of squares of PLS weights, with scores larger than 1 indicating variables are important to the mode. T-tests were performed for compounds with VIP > 1: compounds with p value of <0.05 were considered significantly differentiated compounds.

Molecular Docking
We used Discovery Studio to predict the docking of small molecule compounds with key proteins. The 3D structure of the compound constructed by ChemOffice software was saved in *mol2 format, and its energy was minimized. The 3D structure of the target protein was downloaded from the PDB data (https://www.rcsb.org/), accessed on 24 November 2022, and Discovery Studio 2020 software was used to perform operations such as water removal and hydrogenation on the protein and generate an effective single 3D conformation by minimizing the energy.

Conclusions
In this study, chemical compounds in CR and CC were analyzed and identified using UPLC-Oribtrap-Exploris-120-MS/MS: a total of 58 chemical components were identified. Unsupervised PCA and supervised OPLS-DA were used to assess the differences between CR and CC, and GraphPad Prism 7.0 was used to perform t-tests on the differential components. A total of 26 statistically significant differential compounds were obtained, in which the peak areas of flavonoids (19, 27, 32, 33, 38, 40 and 41) and phenylpropanoids (16, 41, 42, 45, 46, 53 and 58) except cinnamaldehyde in CR were higher than in CC, and the peak areas of terpenoids (35 and 36) and organic acids (2, 3, 8 and 17) in CC were higher than in CR. Additionally, HPLC was used to determine the concentrations and differentiating capacities of coumarin, cinnamyl alcohol, cinnamic acid, 2-methoxycinnamic acid and cinnamaldehyde, which were the major active ingredients in both CR and CC. The results showed that the content of cinnamaldehyde in CC was about twice that in CR, the content of cinnamic acid in CC was similar to that in CR, and the content of coumarin, cinnamyl alcohol and 2-methoxycinnamic acid in CR were significantly higher than that in CC. These five major components could be used as markers for successfully distinguishing CR and CC [63]. Compared to the reported method in the literature, the quantitative method was simpler, and demonstrated good stability and reproducibility. In addition, the different screened constituents were docked to HbA1c, PGC1α, SIRT3 and AMPK, and the results showed that the special and high-concentration components in CR showed high docking scores of affinities with targets such as HbA1c or proteins in the AMPK-PGC1-SIRT3 signaling pathway, suggesting the greater potentials of CR in treating DPN than of CC. Furthermore, the compounds with higher protein binding energy were phenylpropanoid (1, 16, 42, 45) and flavonoids (19,26,27,40), from which it be inferred that flavonoids and phenylpropanoids might be an important material basis for the differential efficacies of CR and CC. The above results provide comparative information on the chemical profiles of CR and CC, as well as the groundwork for exploring the effective substances in each. This is the first time that the compositional differences in the aqueous extracts of CR and CC by LC-MS have been analyzed, which directly reflects the material basis for their different functions under the usage of decoction in clinical practices. The high-content flavonoids and phenylpropanoids in CR may be the key material basis for dispersing wind and cold medicines, and terpenoids and organic acids may be the main active constituents for interior-warming medicines. In previous studies, flavonoids such as quercetin decreased blood glucose levels [64], phenylpropanoids inhibited platelet aggregation [65], and coumarin had suppressive effects on neuropathic cold allodynia in rats [66]. A further investigation of these CR differential substances may make it possible to find effective drugs for treating DPN. The previous studies were mostly focused on trans-cinnamaldehyde, while the CR and CC differential components would be easily acquired, combined with modern separation technology, and are worthy of further modern pharmacological research.  Figure S9: 2D PCA score scatter plot for CR CC and QC after eliminating outliers; Figure S10: 2D OPLS-DA score scatter plot for CR and CC; Figure S11: Validate model of OPLS-DA. Figure S12: Variable importance of projection (VIP) plot of the OPLS-DA. Figure S13: HPLC chromatograms for quantitative analyses. Figure S14: Dendrograms of hierarchical cluster analysis (HCA) for CR and CC.; Figure S15: The workflow of compound screening with Compound Discoverer 3.2. Table S1: Optimization of the LC-MS conditions; Table S2: Compound information with VIP > 1; Table S3: Results of the recovery test; Table S4: Molecular docking fraction; Table S5: Information on CR and CC. Sections S1-S7: Identification of other constituents of CR and CC. References [67][68][69][70][71][72][73]